Superfluids in Flatland Topology, Defects, and the 2016 Nobel Prize in Physics Sid Parameswaran Saturday Mornings of Theoretical Physics Oxford, October 28, 2017 Image Credit: Wikipedia
Superfluids in Flatland By the early 70s, experimentalists were able to create thin films of helium moving on some substrate. Experiments measured deviations in the mass of helium stuck to the substrate, indicative of superfluidity. (some of the helium has frictionless flow!) This contradicts conventional theoretical wisdom that superfluids can t exist in 2D. This conventional wisdom, why it needs to be amended, and the eventual explanation of 2D superfluidity, are the subjects of this talk. [M. Chester, L. C. Yang, and J. B. Stephens Phys. Rev. Lett. 29, 211 (1972)]
Sure it works in practice, but does it work in theory?
Superfluids Macroscopic wave function Superfluid Mass Density Superfluid Velocity Kinetic energy per unit area:
Superfluids Macroscopic wave function Superfluid Mass Density Superfluid Velocity Kinetic energy per unit area: Simplification: assume superfluid density is (roughly) constant everywhere
Phase Waves gradual spatial twist" of phase costs very little energy
Phase Waves gradual spatial twist" of phase costs very little energy energy decreases
Phase Waves gradual spatial twist" of phase costs very little energy energy decreases
Phase Waves gradual spatial twist" of phase costs very little energy energy decreases (This is an example of a Goldstone mode, required by general theorems) Similar examples: sound waves in solids, spin waves in magnets,
Fluctuations Phase-wave modes allow low-energy fluctuations away from perfect order ordered state fluctuation spins no longer nicely aligned!
Fluctuations Statistical Mechanics: probability of fluctuations with energy E at temperature T (so, low-energy fluctuations like phase waves are very important)
Fluctuations Statistical Mechanics: probability of fluctuations with energy E at temperature T (so, low-energy fluctuations like phase waves are very important) Such fluctuations can destroy order (usually happens in 1D/2D) [Variously attributed to Hohenberg/Mermin/Wagner/Coleman (c. 1950s), but originally noted by Peierls for 2D solids in 1934!]
Fluctuations in 2D Superfluids probability Phase-wave modes of many different wavelengths can be thermally excited: What s the phase angle between 0, r? 0 r
Fluctuations in 2D Superfluids probability Phase-wave modes of many different wavelengths can be thermally excited: What s the phase angle between 0, r? 0 r
Fluctuations in 2D Superfluids probability Phase-wave modes of many different wavelengths can be thermally excited: What s the phase angle between 0, r? 0 r
Fluctuations in 2D Superfluids probability Phase-wave modes of many different wavelengths can be thermally excited: What s the phase angle between 0, r? 0 r Averaging over possibilities (statistical mechanics) gives no long range order : phases at 0, r not coherent as r
Logical Conclusion: Long-range phase order is impossible, so 2D superfluids shouldn t exist.
But a Puzzle At low temperatures, But, at high temperatures, High-T and low-t don t match, but no ordering occurs. What s going on?
The answer was provided by J. M. Kosterlitz D. J. Thouless (Oxford D. Phil 1966-69) (work described here was done at Birmingham in the 1970s) Image Credits: Brown University; Mary Levin/University of Washington
Aside: In Praise of Procrastination After his D. Phil in particle physics Kosterlitz took a short-term post-doc in Torino. When this ended, he wanted to move to CERN, where all the excitement was. However, he failed to submit his application on time, and so instead ended up in Birmingham for another post-doc. I had been doing long tedious calculations for very little return, and I was getting a bit fed up. So I started walking from office to office asking if anybody had a tractable problem I could work on. I found myself in David Thouless s office where he was telling me all about superfluid helium films, crystals, dislocations, vortices all concepts that were completely new to me. But somehow what he was saying made sense. So I started working on some of the ideas he was throwing out, and basically things worked out. J.M. Kosterlitz
Back to our simplified model
Back to our simplified model Missing a crucial ingredient: vortices
Back to our simplified model Missing a crucial ingredient: vortices Vortex: point around which φ winds by 2π (integer) + - single vortex single anti-vortex
Vortices are topologically stable + + No smooth twist (e.g. like those in a phase-wave) can change the total phase winding around the circle. The quantized charge (winding) is a topological invariant. [recall the quantized vorticity in John Chalker s talk]
Vortices can destroy long-range order phases around a vortex can t be aligned + + (no matter how you try to smoothly rotate them, some arrows will always be misaligned) This is linked to their topological stability.
Better Description of Superfluid (phase waves) + terms describing vortices We make simplifying approximations all the time by ignoring terms. When is it important to keep the vortices? To answer: need to understand if vortices can cause transitions.
Phase Transitions Since superfluid transition is driven by temperature T, we need to look at free energy E: energy S: entropy Thermodynamics: systems seek to minimize F Low temperatures: lowest-energy state wins (e.g. crystalline solid) High temperatures: high-entropy state wins (e.g. gas) So we need to calculate E, S for vortices
Step 1: Vortex Energy Quantized vorticity: +
Step 1: Vortex Energy Quantized vorticity: + Vortex energy: cost of this phase twist a: size of vortex core L: size of system
Step 2: Vortex Entropy Entropy ~ ln (number of configurations) Different vortex configurations different core location +
Step 2: Vortex Entropy Entropy ~ ln (number of configurations) Different vortex configurations different core location +
Step 2: Vortex Entropy Entropy ~ ln (number of configurations) Different vortex configurations different core location +
Step 2: Vortex Entropy Entropy ~ ln (number of configurations) Different vortex configurations different core location +
Step 2: Vortex Entropy Entropy ~ ln (number of configurations) Different vortex configurations different core location number of configurations +
Step 3: Vortex Free Energy Putting all the factors in: Changes sign at
Kosterlitz-Thouless Vortex Transition [J.M. Kosterlitz & D.J. Thouless, J. Phys. C 5, L124 (1973); ibid, 6, 1181 (1973)] vortices are expensive and rare (ignoring them is OK) phase waves lead to vortices are cheap and proliferate (can t be ignored) and lead to No long-range order in either case, so no order parameter Kosterlitz & Thouless called this a topological phase transition.
Coulomb Gas Analogy: vortices/antivortices +/- point charges with logarithmic interactions (useful for quantitative calculations) transition from gas of dipoles to a plasma Tc gas of tightly bound dipoles plasma of unbound charges - + - + - + - + +- + - + - + - + - - - + + + - + - - + + + - - + + - - T
KT Transition: Experimental Signatures Superfluid transitions in 3D can be seen via thermodynamic measurements e.g. specific heat But thermodynamic quantities are smooth across the KT transition! We need a different probe to detect the phase transition. Image From: F. London, Superfluids
KT Transition: Experimental Signatures Simplification: assume superfluid density is (roughly) constant everywhere But: Superfluid wavefunction must vanish in vortex core So as vortices proliferate, superfluid density should decrease + Naïve expectation: ρs vanishes continuously at transition
KT Transition: Experimental Signatures [D. Nelson & J.M. Kosterlitz, PRL 39, 1205 (1977)] Simplification: assume superfluid density is (roughly) constant everywhere But: Superfluid wavefunction must vanish in vortex core So as vortices proliferate, superfluid density should decrease + Actually: Naïve expectation: ρs has a universal ρs vanishes jump continuously at the transition at transition that can be measured universal jump
KT Transition: Experimental Signatures Can measure ρs using torsional oscillator torsion rod, spring constant κ substrate moment of inertia I T>Tc: all the He-4 contributes to I (by sticking to substrate) decrease in Tosc is measure of ρs T<Tc: superfluid fraction decouples, so I, Tosc both decrease [D.J. Bishop, J.D. Reppy, Phys. Rev. B 22, 5171 (1980); additional data on dissipation not shown] [original expt s: E. L. Andronikashvili, Sov. Phys JETP 18, 424 (1948), building on ideas of Landau]
KT Transition: Experimental Signatures + also other probes (a.c. response, ) Bishop & Reppy: measured ρs and found the universal jump predicted by the KT theory universal jump [D.J. Bishop, J.D. Reppy, Phys. Rev. Lett. 40, 1727 (1978)] (Tune thickness of film to change Tc)
Honorable Mentions V.L. Berezinskii Noted that vortices were important in destroying order (~1.5 yrs before Kosterlitz and Thouless!) Understood relation to Coulomb gas + unbinding transition But: predicted a very different transition, e.g. a continuously vanishing superfluid density (1935-1980) Died in 1980 after a long illness. [V.L. Berezinskii, Sov. Phys. JETP 32, 493 (1971); ibid, 34, 610 (1972)] F. Wegner First model(s) for phase transitions w/o long-range order Now understood as confinement-deconfinement transitions in lattice gauge theories (1940- ) Crucial to understanding topological order and quantum spin liquids Image Credits: http://www.edu.delfa.net/ ; University of Heidelberg [F. Wegner, J. Math. Phys. 12, 2259 (1971)]
Variations on a Theme of Kosterlitz and Thouless Topological defects are relevant to systems as diverse as: liquid crystals quantum magnets superfluid 3 He (Tony Leggett s Nobel prize) cosmic strings in the early Universe Verduzzo Lab, Rice Univ.; K/ Everschor-Sitte/M. Sitte; http://ltl.tkk.fi/research/theory/he3.html; http://cp3.irmp.ucl.ac.be/~ringeval/strings.html;
Also impact the big questions of Condensed Matter Can we exhaustively classify all phases of matter? Can we understand all possible phase transitions?
The Landau Paradigm distinct phases of matter = distinct long-range order patterns phase transitions = changes of these patterns Lev Landau crystals magnets superfluids Image Credits: Nobel Foundation; Didier Descouens; Eurico Zimbres ; Alfred Leitner.
Beyond The Landau Paradigm 2D superfluids don t really have long-range order But there is a distinct superfluid phase with power-law phase correlations. Kosterlitz & Thouless discovered a continuous phase transition that falls outside the Landau paradigm. In essence, by replacing symmetry with topology as a guiding principle, we can go beyond the Landau paradigm. We re still discovering new surprises today.
David J. Thouless, F. Duncan M. Haldane and J. Michael Kosterlitz were awarded the 2016 Nobel Prize in Physics for theoretical discoveries of topological phase transitions and topological phases of matter" They also contributed many other topological ideas to quantum condensed matter including work that led to the discovery of topological insulators. Image Credits: Brown University; Bengt Nyman; Mary Levin/University of Washington